Visualize port scan results in a self-contained HTML file.
In short: We interpret a host as a point in a vector space with 2^17 dimensions over F_2. Each dimension corresponds to a TCP- or UDP-port and has either value 0 or 1, depending on its state. Then we apply dimensionality reduction techniques to project the data onto two dimensions.
Each circle in the plot corresponds to one group of hosts. The size of the circle is related to the size of the group. Hosts with the same port configuration are grouped together. Similar groups should be close by. The colors mean nothing - except for gray: no open ports. The coordinates are also not meaningful and can change with a new run.
If you require instructions on how to install a standard Python package, I
recommend you use uv:
# Using uv (recommended):
uv tool install git+https://github.com/SySS-Research/Scanscope.git
# Alterantively, using pipx:
pipx install git+https://github.com/SySS-Research/Scanscope.gitUnfortunately, the requirements (in particular the machine learning
dependencies including numpy and pandas) are quite heavy with almost
600MB, so be prepared.
Convert nmap output in XML to HTML:
scanscope nmap_output.xml -o scanscope.htmlHint: The more ports you scan, the better this should work.
I recommend scanning at least the top 100 ports, so: nmap -T4 -sS -F -oX nmap_output.xml -iL input.txt. Service scans or script scans do not help.
Scanning the top 1000 ports or even all ports however, does.
For more infomation, read the output of scanscope -h.
For advanced users who want to tune the parameters for optimal visualization quality, an optional optimization module is available. This uses Bayesian optimization to automatically find the best parameters for your specific dataset.
Install the optimization dependencies:
$ uv sync --group optimize
Run the optimization:
$ uv run examples/optimize_umap.py nmap_output.xml --trials 50
See examples/README.md for detailed usage instructions.
MIT licensed, developed by Adrian Vollmer, SySS GmbH.